The AI agent course for beginners — go from zero to building, deploying and selling AI agents.
16 modules, taught in Urdu and English. You build one real AI Employee from an empty folder to a live URL — then your own. No coding background needed. $100 one-time, lifetime access. Updated September 2026.
Pay directly — bank transfer · JazzCash · Easypaisa
Every module adds one piece. By Module 13 you have all five, live, on a URL you can send to anyone.
Watch what you'll be building — running my business today.
The AI Employee that handles DSP's admissions — Zara, our sales AI Employee — the content system behind 6 million views, and what the program looks like from the inside. Recorded by Sardar, not a marketing team.
What is this AI agent course, in one paragraph?
DSP AI Agent Mastery is a self-paced AI agent course for complete beginners. Across 16 modules you build one working AI Employee — a café ordering agent — from an empty folder to a live URL using Claude Code, then learn to sell agents like it to real businesses. Lectures are in Urdu and English. $100, lifetime access.
| Level | Complete beginner — no coding |
|---|---|
| Format | Self-paced video, 30+ hours, 16 modules |
| Language | Urdu–English lectures, English materials, subtitles |
| Project | One real AI Employee → live URL, plus your own capstone |
| Certificates | DSP Master certificate (verifiable URL) + 3 Claude Academy badges from Anthropic |
| Price | $100 one-time · bank transfer, JazzCash, Easypaisa · card on request |
Can a beginner with no coding background really build an AI agent?
Yes. You describe what you want in plain words and Claude Code writes the code; your job is to plan, direct, test and ship, which is exactly what the 16 modules teach. If you can write a clear WhatsApp message, you can write the Job Description that runs an agent. That is the whole gap this program closes — and it is a smaller gap than it looks from the outside.
You already use AI every day. You still can't build with it. Here is why, and what changes.
Chatting, not building
You ask Claude or ChatGPT a question, copy the answer, and start again tomorrow. Nothing you made on Monday is still working for you on Friday.
Every tutorial assumes you code
Free material jumps from "what is an agent" straight to Python and API keys, so you stop at the first error nobody explains.
You describe it — then you ship it
Sixteen modules, one build at a time, ending in a live URL that keeps running after you close the laptop, and a system you can sell to someone else.
What will I actually build in this AI agent course?
A café ordering AI Employee. It starts as an empty folder in Module 4, becomes a four-page website, then an agent that takes orders, remembers customers, emails the kitchen and writes to Google Sheets through MCP — built with Claude Code, versioned on GitHub, deployed on Vercel to a public URL, and by Module 14 serving two cafés from one codebase.
Every module adds one real capability to it. It is a simplified version of the multi-tenant restaurant agent DSP runs for clients — the same shape, built by you.
- M04–05WebsiteHome, menu, about, order form. Responsive. No code written by hand.
- M07Ordering agentTakes an order item by item, handles corrections, confirms the total.
- M08–10ConnectedClaude API in the backend, orders emailed to the kitchen, written to Sheets via MCP.
- M09MemoryKnows the menu and allergens. Remembers a returning customer's last order.
- M11–13LiveTested with the 10-Question sheet, hardened against prompt injection, deployed on Vercel.
- M14Two cafés, one systemMulti-tenant. This is the shape of what you sell to clients.
Who is this AI agent course for (and who is it not for)?
It is for people who use Claude or ChatGPT every day and want something that keeps working after they close the laptop — students, teachers, freelancers, marketers and business owners, in Pakistan or abroad. It is not for you if you want LangGraph internals, framework theory or a computer-science course: this is a builder's course, and Claude Code writes the code.
If you can describe what you want in plain words, you can build an agent. That is the entire skill this program teaches.
Skills a job description asks for
Leave with a live agent on GitHub and a certificate that verifies what you built — not what you watched.
Automate the admin, keep the teaching
Build an agent that answers parent queries, grades against a rubric, or drafts lesson plans from your notes.
Add a $500–$2,000 service line
Deliver agents to clients using the same discovery, proposal and pricing templates DSP uses. See real client results and DSP's own pricing for done-for-you agents to calibrate yours.
Agents that run your pipeline
Lead qualification, follow-up sequences, content research — built by you, connected to your tools.
Your first AI employee
A booking, ordering or support agent for your own business, running on WhatsApp, web or email — the way Zara, our sales AI Employee, runs DSP's admissions.
Learn in Urdu and English, on your own time
Lectures are taught in an Urdu–English mix, with English materials and subtitles. Study at 6 am in Dubai or midnight in Manchester. Nothing is live-only.
What does the 16-module AI agent curriculum cover?
Five phases. Phase 0 (Modules 1–3) gives you the foundations: what an agent is, how to write a Job Description, and how to set up Claude, ChatGPT and Gemini. Phase 1 (4–6) teaches vibe coding with Claude Code, websites and Git. Phase 2 (7–10) turns your website into an agent with APIs, RAG, memory and MCP. Phase 3 (11–13) tests, secures and deploys it. Phase 4 (14–16 plus the capstone) makes it multi-tenant, teaches you to sell it, and adds the AI Search Dominance Engine.
Modules open in order, one build at a time — that's how everyone finishes. No tests, no grades. You watch, you build, you move on.
Zero — Foundations
You produce: one chosen agent idea, a tested 7-Part Job Description, and a working Claude Project.
Builder
You produce: AI Employee Parts 1–2 — a responsive café website, on GitHub, from an empty folder.
Agent Engineer
You produce: AI Employee Parts 3–6 — takes orders, remembers customers, emails the kitchen, writes to Google Sheets.
Production
You produce: AI Employee LIVE — a public URL you can send to anyone.
Seller
You produce: your AI Employee serving two cafés from one system, a real client proposal, one SEO page built by the AI Search Dominance Engine, and your own original agent — the capstone that earns the certificate.
How does a self-paced AI agent course work day to day?
You watch one chaptered lecture of 35–60 minutes, build that module's project with its template and copy-paste prompts, post your screenshot or URL in the DSP group, and mark the module complete so the next one opens. About an hour a day finishes in 30 days; four sessions a week finishes in eight. No tests, no grades — and if you get stuck, the weekend live session debugs your build on the call.
Watch. Build. Share. Move on. The same rhythm every module, so you never wonder what to do next.
Chaptered lectures, 35–60 minutes, taught in Urdu and English.
A specific project with a template and copy-paste prompts. Always a real artefact.
Post your screenshot or URL in the DSP group. Stuck? Ask — someone ahead of you has hit it.
Mark the module complete and the next one opens. Badges at each phase.
What certificates do I get, and are they recognised?
Two kinds. The DSP Master certificate has a public verification page showing the live agent you built — proof of work, not attendance. Module 3 also takes you through Claude Academy, Anthropic's free training, so you finish holding three of their completion badges with your name on them. Neither is a university accreditation; what employers and clients check is the working agent behind the link.
Three certificates from Claude — Anthropic, USA. And one from DSP. Every certificate has a public verification page showing the capstone URL and repo behind it. An employer or client can open it and see the agent working. Featured capstones earn Master with Distinction.
You finish with four, and three of them are not ours. Module 3 takes you through Claude Academy, the training run by Anthropic — the American company in San Francisco that builds Claude — and you come out holding Claude 101, Claude Code 101 and Introduction to Claude Cowork, each with your own name on it. Anthropic issues them, so they say nothing about DSP and everything about you. The fourth is the DSP Master certificate, and it is the one that points at a live agent you built.
None of this is out of reach. Every DSP student who follows the module earns all three — no exam fee, no waiting list, no degree required. Ours have been earning them for months, and we publish every one on our channels the week it lands. How the Anthropic certificates actually work — who issues them, and why no institute can — is set out in the Claude Code course guide.
What have students actually built with this course?
Real builds, on camera. Mohsin, a finance professional in the UK with no software background, built his first website and an AI agent in the program and deployed it live. A student in Pakistan earned PKR 60,000 from AI work before he had finished. Every story below is a lesson inside Module 15, told by the student — not a written quote.
From the live cohorts this program was recorded in. More builds are added as students give permission to name them.
First website + AI agent, deployed liveA finance professional with no software background. He built his first website and AI agent in the program and deployed it live.
PKR 60,000 of paid AI workEarned PKR 60,000 from AI work before he had even finished the bootcamp.
Agents, automation and prompt engineering, applied at workCame for practical skills — AI agents, automation and prompt engineering — and says the training changed how she works.
Results depend on the work you put in. These are individual students, not a promise of income.
What does it cost to run an AI Employee after you build it?
Less than most people expect. A deployed AI Employee has three running bills: the Claude API, which is metered per token and is the only cost that grows with orders; hosting, which is free on Vercel's Hobby plan for what this course deploys; and a messaging channel such as WhatsApp, which is optional and billed per conversation by Meta. GitHub and Google Sheets are free.
Here is every bill the café agent can generate, who sends it, and where it is free. Module 15 shows how to price an agent against these costs so a client's monthly fee covers them with margin — the same calculator is in the Client Acquisition Kit.
| What | Billed by | Free tier | Where the course covers it |
|---|---|---|---|
| Claude API | Anthropic, metered per token | No monthly minimum — you pay for what the agent uses | The only bill that grows with orders. Module 8 covers handling cost; Module 12 caps what the agent may spend. |
| Hosting — website + agent backend | Vercel | Hobby plan is free and covers what this course deploys | A client's commercial traffic belongs on Vercel's paid plan, which starts at $20 a month. |
| Code + version control | GitHub | Free | Public or private repo, unlimited commits — Module 6. |
| Kitchen email + Google Sheets via MCP | Free with a Google account | Orders are emailed and written to Sheets through MCP — Modules 8 and 10. | |
| WhatsApp or voice channel | Meta (per conversation) or a telephony provider | Optional — the café agent in the course runs on a web URL | Zero until you add a channel. Modules 14 and 15 cover when a client's use case justifies one. |
Who teaches this AI agent course?
Sardar Ghaffar, founder of Digital Services Program: 24 years teaching IT in London, the UAE and Pakistan, now running the agency that builds the AI Employees DSP builds for clients. He is a Google-verified AI Agentic Trainer and an Anthropic (Claude)-verified educator, and he recorded every lecture in this course himself.
Sardar Ghaffar
I was in IT before most of my students had email. Twenty-four years of teaching — in London, the UAE and Pakistan — and the last few spent on one thing: showing people who aren't programmers how to build AI agents that work and earn.
Everything in this program is what I use with real clients. The templates are the ones my team uses. The AI Employee you build is a simplified version of the multi-tenant agents DSP builds for businesses. I've watched thousands of students go from "what is an agent?" to a live URL. This is the path they took, cleaned up.
Read more about Sardar Ghaffar, including the public verification links for each credential. He also teaches the ChannelOps course for people who run a YouTube channel.
What's included for $100?
Everything you need, nothing you'll never open.
- 30+ hours of lectures, cut and chaptered, across 16 modules~30 hrs
- AI Employee build thread — 8 parts, empty folder to two live cafés1 project
- Original capstone, reviewed by the DSP team1 project
- Resource Vault — 100+ prompts, 25 starter Job Descriptions, templates, checklists100+ files
- AI Employee build-prompts PDF, copy-paste readyPDF
- Client Acquisition Kit — discovery sheet, proposal, pricing calculator, contract, outreach packbonus
- One Agent Many Clients blueprint — the multi-tenant model DSP sellsbonus
- One year of free support in the DSP group12 months
- Live weekend session — bring your build, we debug it on the call, recordedevery weekend
- DSP Master certificate with a verifiable URL + 4 milestone badgeson completion
- Claude Academy badges: Claude 101, Claude Code 101, Claude Coworkissued by Anthropic, USA
- Lifetime access and every future updatelifetime
Sixteen modules. Every one ends with something you built.
Around 30 hours of lectures, recorded live with real students asking real questions — cut, chaptered and put in the order a beginner needs. Every module, its outcome and what you build, on one page: the full curriculum.
M01AI Foundations+
M02Prompting & Context Engineering+
M03Claude · ChatGPT · Gemini+
M04Vibe Coding+
M05Websites+
M06Git & GitHub+
M07AI Agents+
M08APIs+
M09RAG & Memory+
M10MCP+
M11Testing & Observability+
M12Security+
M13Deployment+
M14Multi-Agent & Business Automation+
M15Selling AI Solutions+
M16AI Search Dominance Engine+
CAPCapstone — your own agent+
How does this compare to other AI agent courses for beginners?
There are excellent free and university options, and for some people they are the better choice. Here is where each fits: Mastery is the one built for non-coders, that ends in one deployed multi-tenant project, that teaches you to sell it, and that you can pay for from Pakistan in Urdu-taught lectures for $100 once.
| DSP AI Agent Mastery | Udemy / Coursera courses | University certificate | Free open-source (Microsoft, Hugging Face) | YouTube | |
|---|---|---|---|---|---|
| Built for non-coders | Yes — Claude Code writes the code | Mixed | Usually assumes Python | Developer-first | Mixed |
| One complete project to a live URL | Yes — the café AI Employee, multi-tenant | Rarely | Sometimes | Notebooks, not deployed | No |
| Teaches selling agents to clients | Yes — Phase 4 plus the Client Acquisition Kit | No | No | No | No |
| Language | Urdu + English lectures, English materials | English | English | English | Varies |
| Human support | 1 year in the DSP group, weekend live debug sessions | Q&A board | Cohort TA | Community forum | None |
| Certificates | DSP verifiable certificate + 3 Anthropic badges | Platform certificate | University credential | Varies — Hugging Face issues a certificate | None |
| Pay from Pakistan | Bank transfer, JazzCash, Easypaisa | Card | Card, USD | Free | Free |
| Price | $100 once | Typically $15–$200 | Paid, USD | Free | Free |
One more option that no longer exists: the DSP Agentic Lab bootcamp, the live seven-day cohort these lectures were recorded in, closed in August 2026 after six batches. Mastery is its self-paced successor — same curriculum, same café agent, start any day.
Can I learn to build AI agents from Pakistan, in Urdu?
Yes. As far as we know this is the only AI agent course for beginners taught in an Urdu–English mix, priced at $100, and payable by bank transfer, JazzCash or Easypaisa. Lectures were recorded in DSP's live cohorts, taught from Islamabad; templates and slides are in English so your work is client-ready anywhere. DSP has taught students in the UK, UAE, USA, Canada and Pakistan this way.
یہ کورس ان لوگوں کے لیے ہے جو روز AI استعمال کرتے ہیں مگر خود AI ایجنٹ بنانا نہیں جانتے۔ سولہ ماڈیولز، اردو اور انگریزی میں لیکچرز، ایک مکمل AI ایمپلائی جو آپ خالی فولڈر سے لائیو URL تک خود بناتے ہیں — اور پھر اپنا ایجنٹ۔ فیس صرف $100، لائف ٹائم رسائی، ایک سال سپورٹ۔
Pay in PKR at the bank details on the enrol page, upload the screenshot, and your sign-in link arrives once it is verified — usually within a few hours. Study at 6 am in Dubai or midnight in Manchester; nothing is live-only. How the Urdu–English teaching works, module by module: AI Agent Course in Urdu. Joining from the Gulf, the UK or North America: the guide for overseas Pakistanis.
Read six free lessons first.
Each post below is the idea behind one module, written for a complete beginner. If they make sense, the course will too.
AI agent course FAQ
Before you ask.
I have never coded. Can I really do this?
Yes. That is the audience this was built for. You describe what you want and Claude Code writes the code. Your job is to plan, direct, test and ship — which is what the program teaches. If you can write a clear WhatsApp message, you can write a Job Description.
How much time does it take?
About an hour a day for 30 days, or four sessions a week for eight weeks. Each module is one to three lectures plus a build. Lifetime access means you can go slower — the only thing that doesn't work is stopping.
Windows or Mac?
Both. Setup guides for each are in Module 3, including the fixes for the common Windows issues we've solved across 338 students.
Is $100 really all I pay?
$100 covers the entire program, lifetime. You will need a Claude account to build with during the modules — Module 3 walks you through setup and the free-tier route, and paid Claude plans start at $20/month if you choose to upgrade. GitHub and Vercel are free for what this program uses. No other purchase is required to finish.
What does "one year of free support" mean?
You're in the DSP group from day one, and questions there are answered within one working day. On top of that there's a live session every weekend — bring your build, we debug your errors on the call, and every session is recorded. Support runs for a full year.
Is it in Urdu or English?
Lectures are taught in an Urdu–English mix, the way DSP teaches live. All templates, slides and downloads are in English. Subtitles are provided.
Can I pay from Pakistan?
Yes — bank transfer, JazzCash or Easypaisa. Go to the enrol page, send the payment, upload the screenshot with your email, and we send your sign-in link once it's verified — usually within a few hours.
Do I get a recognised certificate?
Two kinds. The DSP Master certificate has a public verification page showing the live agent you built — that is proof of work, not attendance. Separately, Module 3 walks you through Claude Academy, the free training run by Anthropic — the US company that builds Claude — so you finish holding three of their course completion badges with your name on them — Claude 101, Claude Code 101 and Introduction to Claude Cowork. Those are issued by Anthropic, not by DSP. Neither is a university accreditation; what employers and clients actually check is the working agent behind the link.
Refunds?
Seven days, no questions. Start Module 1; if it isn't for you, email us and we refund in full.
Is this the right AI agent course for beginners, or should I start with Python?
You don't need Python. Module 4 (Vibe Coding) teaches you to direct Claude Code: you plan a feature, describe it, read and test the code it writes, commit, and move to the next one. You learn to read code long before you ever write it by hand — and most students never need to.
Which tools does the course use?
Claude, Claude Code, ChatGPT and Gemini (set up in Module 3), GitHub and Vercel (Modules 6 and 13), MCP servers (Module 10) and a vector store for RAG (Module 9). Everything runs on a free tier or costs under $20 a month; the only paid upgrade anyone chooses is a Claude plan.
How is this different from the DSP live bootcamp?
Same curriculum, same café AI Employee. The DSP Agentic Lab bootcamp was the live, cohort-based version and closed in August 2026 after six batches; these lectures were recorded in those cohorts. Mastery is the self-paced version: recorded lectures you start any day, plus a live weekend support session and the DSP group for a year.
Can I build an agent for my own business instead of the café?
Yes. The café agent is the guided build — every module adds one capability to it so you are never guessing what to do next. Your capstone is your own agent for your own use case, built with the same method, and it is reviewed by the DSP team before the certificate is issued.
What jobs or income can this lead to?
Three routes we see most: freelance agent builds for local businesses, monthly automation retainers, or an AI Employee running inside your own business. Phase 4 includes the discovery sheet, pricing calculator, proposal and contract templates DSP uses with its own clients. Results depend on the work you put in; nothing here is a promise of income.
Can I learn to build AI agents without knowing English well?
Yes. The lectures explain everything in Urdu; only the tool names, prompts and code are in English, and the course gives you those as copy-paste templates. If you can read a WhatsApp message in English, you can follow this course. Subtitles are provided on every lecture.
Do I need a powerful laptop?
No. Any Windows or Mac machine from roughly the last six years with 8 GB of RAM is enough. Claude Code runs in a terminal, the model runs in Anthropic's cloud, and your agent is hosted on Vercel — the heavy work never happens on your machine.
Module 1, Lesson 1 is waiting.
One hundred dollars, once. Lifetime access. A year of support. A live agent with your name on it at the end.
digitalservicesprogram.com · +92 342 0580864 · Office 151, Plot 7, 10th Avenue Service Road East, Sector F-10/4, F-10 Markaz, Islamabad 44000, Pakistan